Problems in Replica Server Placement (RSP) over Content Delivery Networks (CDN)
Darothi Sarkar, Nitin Rakesh, K. K. Mishra · 2015
Massive traffic generated by a huge number of requests is the major concern in today's web based world. Content Delivery Network (CDN) was evolved to cope up with this issue which is a popular research area in recent years. In CDN all the content are stored in a server which is known as the origin server. The major challenge in CDN is to place a number of mirror images i.e. replicas of the origin server at the edge of the Internet. In this paper a comparative analysis is performed on replica placement algorithms like Tree-based, Greedy, Random, HotSpot, HotZone as well as some recently evolved strategies like GeoIP, Flow Count, NetClust. In this paper we have formulated a problem and a comparative analysis on the above mentioned replica placement algorithms are performed based on complexity, optimization factors and their applicability. The comparison reveals that Greedy provides placement solution which is almost optimal but its complexity is a hindrance to its implementation. HotSpot becomes very popular from implementation point of view. Because of the reduced cross bandwidth traffic the Flow Count strategy attracts the video generating companies. NetClust proves its performance to be the best as its complexity, deployment cost and delay are lower than all other approaches discussed in this paper. The analysis indicates that though different approaches are having their own merits and demerits, but their use is application specific. Selection of replica placement approaches depends on the designer and the types of applications.